Platform modules
AI engine modules
Four composable engines: blood marker extraction, image signal analysis, progression scoring, and structured reporting. Deploy individually or as a pipeline.
- PDF lab reports (digital + scanned)
- Scalp and hair imagery (JPEG, PNG)
- Structured intake metadata (JSON)
- Reference ranges and units config
- Native PDF text extraction
- OCR pipeline for image-based PDFs
- Vision provider integration
- Canonical marker normalisation
- Signal vector computation (0–1)
- Structured JSON (markers + signals)
- Domain scores and risk tiers
- Explainability vectors
- PDF reports with audit trail
- REST API response payloads
Blood Marker Intelligence Engine
- PDF and image lab report ingestion with multi-format support
- Structured biomarker extraction with units, reference ranges, and flags
- Confidence scoring per marker and aggregate confidence per report
- OCR fallback for scanned/image-based documents
- Normalisation to canonical marker names for downstream processing
Parses lab outputs into structured JSON. Handles digital PDFs natively; falls back to OCR for scanned documents. Output schema includes marker name, value, unit, reference range, flag (low/normal/high), and per-marker confidence. Designed for batch processing and integration with LIMS.
Image Signal Extraction Engine
- Scalp and hair imagery analysis with configurable resolution
- Extraction of visibility, redness, flaking, and lighting proxies
- Blur and quality assessment for input validation
- Donor-pattern and texture uniformity estimation
- Confidence-weighted outputs with summary captions
Consumes image bytes and returns normalised signal vectors (0–1). Supports heuristic and vision-provider backends. Outputs include scalar proxies for scalp visibility, redness, flaking, lighting, and blur. Used for longitudinal tracking and as input to progression models.
Progression Velocity Engine
- Five-domain scoring: androgen, inflammation, thyroid/metabolic, nutrients, stress
- Configurable weights and risk tier thresholds
- Explainability vectors per domain with driver attribution
- Integration of blood markers and image signals
- Overall score aggregation with tier classification (low/moderate/elevated/high)
Combines blood marker and image signal inputs into domain-level scores. Each domain uses flag-based and range-based logic with tunable weights. Produces explainability strings for auditability. Output includes domain_scores, overall_score, risk_tier, and explainability map.
Structured Reporting Engine
- PDF report generation with configurable templates
- Version control and audit trail per report
- Approval workflow (draft → approved → released)
- Signed URL generation for secure distribution
- Support for custom branding and layout
Renders premium PDFs from scoring outputs and intake metadata. Reports follow a defined lifecycle with status transitions. Storage paths and versioning are tracked for compliance. Supports multi-tenant deployment with tenant-specific templates.